Bovine tuberculosis (bTB) is an important agricultural disease from both an economic and a public health perspective. Mycobacterium bovis, the cause of bTB has a wide host range including cattle, white-tailed deer (Odocoileus virginianus), and humans1, and causes decreased milk and meat production in cattle. Humans can become infected by contact with infected animals or, more commonly through consumption of unpasteurized milk2. Risk of zoonotic bTB infection has been greatly reduced in developed countries by a combination of effective test and slaughter control programs and milk pasteurization; however, M. bovis is still responsible for 1-2% of human tuberculosis cases in the US3. Due to the risk of zoonotic infection, and the production losses caused by bTB, the US instituted a bTB eradication program in 1917. This program includes slaughter surveillance, and a lengthy testing protocol for infected herds estimated to cost over $1.5 million in a 1000 cow dairy herd4. Despite over 100 years of directed national control, outbreaks still occur in cattle every year.

Due to the potential lag time between infection and detection, as well as the multiple possible sources of infection, identifying the source of infection in a herd is difficult. Although thorough sampling is routine in traceback investigations, animal movement records are often incomplete, and other potential sources of infection, including zoonotic transmission, are difficult to characterize. To maximize the information available from outbreak investigations, whole genome sequencing (WGS) of M. bovis isolates from infected herds was incorporated into the bTB eradication program by the NVSL in 20138. Using the over 4500 M. bovis isolates collected, maintained and sequenced by the National Veterinary Services and NVSL, we will fill a critical gap in determining the source of an outbreak by predicting time and location of infection. Our objective is to utilize M. bovis WGS data to increase efficiency in outbreak investigations by 1) estimating time since herd infection and investigating drivers of variation in M. bovis evolutionary rate, and 2) predicting the geographical source of outbreaks by analyzing pan-genome population structure.

Our overall goal is to narrow down the source of a bTB outbreak in both time and space. To accomplish this goal, we will create two classes of models: Bayesian phylogenetic models used estimate evolutionary rate and time since infection, and population genetic structure models to estimate the geographical source of an outbreak. We will estimate not only M. bovis evolutionary rate, but also drivers of variation in evolutionary rate, and we will use these estimates to create a model that can be used prospectively to determine time since infection and aid in traceback. The population genetic structure model developed will be capable of predicting the likely source of an M. bovis outbreak, which can help direct the traceback investigation to the correct place. We will also produce a dynamic M. bovis database that can also be expanded to include sequences from new outbreaks to create a global network of M. bovis geolocations.